Most coverage of Cobenfy’s move into Alzheimer’s disease treats it as a single effort, and it is not. Bristol Myers Squibb is running two separate Phase 3 programs in Alzheimer’s, for two distinct regulatory indications, and they are very different commercial propositions. The one that gets the attention is ADEPT, which targets psychosis associated with Alzheimer’s and has pivotal readouts due by the end of 2026. The quieter one is ADAGIO, which targets agitation, and it matters precisely because agitation, unlike psychosis, already has an approved drug. That changes what Cobenfy has to prove.
Two programs, two indications
The distinction is not cosmetic. Psychosis and agitation are separate neuropsychiatric symptoms of dementia, regulated as separate indications, measured on different scales, and tested in different trials. The ADEPT program, comprising ADEPT-1, ADEPT-2, and ADEPT-4, evaluates Cobenfy against the hallucinations and delusions of Alzheimer’s psychosis, with the Neuropsychiatric Inventory measure of hallucinations and delusions as the primary endpoint. The ADAGIO program, ADAGIO-1 and ADAGIO-2, evaluates it against agitation, using the Cohen-Mansfield Agitation Inventory as the primary outcome. The agitation trials each plan to enroll around 352 biomarker-confirmed Alzheimer’s patients with a history of agitation and aggression, treat them for roughly 14 weeks, and they began in July 2025 with completion planned for late 2028. Cobenfy here is paired with an extended-release companion formulation. The two programs run on entirely different clocks: psychosis reads out this year, agitation not until the end of the decade.
The part that changes the bet
In psychosis, there is no FDA-approved treatment specific to Alzheimer’s, which is why ADEPT is framed as a potential first-in-class win. Agitation is different. In May 2023 the FDA approved Rexulti, brexpiprazole, as the first and only pharmacological treatment for agitation associated with Alzheimer’s dementia, on the strength of Phase 3 trials showing roughly a 31 percent greater reduction in agitation symptoms than placebo on the Cohen-Mansfield inventory. So Cobenfy is not entering an empty field. It is entering as a challenger to an incumbent that has been on the market for three years.
That reframes the entire pitch. Against no competitor, Cobenfy needs only to work. Against Rexulti, it needs to differentiate, and the differentiation it would lean on is safety. Rexulti is an atypical antipsychotic, and it carries the class boxed warning for increased mortality in elderly patients with dementia. Cobenfy is a muscarinic M1 and M4 agonist that does not block the dopamine D2 receptor, which is the basis for positioning it as a non-antipsychotic alternative. The implicit argument is that a drug without the dopaminergic mechanism might match the incumbent’s efficacy while avoiding the safety baggage that the only approved option carries. Bristol Myers has already signaled it will lean on safety as the key differentiator in its Alzheimer’s work, and in agitation that argument has a specific target rather than an open field.
Why this is the harder of the two bets
The agitation program is the more difficult commercial story for several reasons, and they compound.
Agitation trials in dementia are notoriously placebo-heavy. The improvement seen in placebo arms is large, in part because the structure and attention of a clinical trial themselves reduce agitation, which makes a drug-placebo separation hard to produce and easy to lose. Even the approved incumbent’s effect, while real, is modest against that backdrop. A muscarinic agent has to clear the same noisy bar.
Cobenfy’s recent record also argues for caution about extrapolating its schizophrenia success into new populations. Its adjunctive schizophrenia trial missed, and its lead Alzheimer’s psychosis study, ADEPT-2, was delayed after the company identified conduct irregularities at some sites and had to enroll additional patients. Neither bears directly on agitation, but both temper the assumption that the mechanism travels cleanly.
The safety pitch, finally, has to survive contact with the patient population. Cobenfy’s tolerability burden is cholinergic, including gastrointestinal effects, and frail elderly dementia patients are exactly the group in whom tolerability problems matter most. A safety-based challenge to Rexulti only works if Cobenfy’s own side-effect profile holds up in that population at scale, which the trials have yet to show.
And the timeline is unforgiving. ADAGIO completes around 2028, years after the psychosis readout and five years after Rexulti’s approval. The incumbent will have a long head start and an established prescriber base by the time any agitation data arrives.
The frame
Read correctly, Cobenfy’s Alzheimer’s strategy is two bets on the broad opportunity of neuropsychiatric symptoms in a population of roughly 7 million Alzheimer’s patients, of whom about half experience agitation and a large share experience psychosis. The psychosis bet, ADEPT, is the near-term, first-mover play into an unserved indication, and it is the catalyst worth watching this year. The agitation bet, ADAGIO, is the longer-dated, harder play: a challenge to an entrenched if imperfect incumbent, on a placebo-heavy endpoint, with data that will not arrive until the end of the decade. The two are routinely collapsed into a single “Cobenfy for Alzheimer’s” story. They are different races, and the agitation one is the steeper climb.